International Journal of Trend in Scientific

Research and Development (IJTSRD)

International Open Access Journal
ISSN No: 2456 - 6470 | www.ijtsrd.com | Volume - 2 | Issue – 1

Elementary approach
pproach towards
owards Biological Data Mining

Faiz Hashmi
Department of Biotechnology, IILM Academy of
Higher Learning, Greater Noida, Uttar Pradesh, India

ABSTRACT
In this paper we provide an overview on interactive and
integrative knowledge discovery and data mining. The
most important challenges, includes the need to develop
and apply novel methods, algorithms and tools for the
integration, fusion, pre-processing,
processing, mapping, analysis
and interpretation of complex biomedical data with the
aim to identify testable hypotheses, and build real
realistic
models. The HCI-KDD KDD approach, which is a
synergistic combination of methodologies and
approaches of two areas, Human–Computer
Computer Interaction
(HCI) and Knowledge Discovery & Data Mining
(KDD), offer ideal conditions towards solving these
challenges: with the goal of supporting human
intelligence with machine intelligence. There is an
urgent need for integrative and interactive machine
learning solutions, because no medical doctor or
biomedical researcher can keep pace today with the
increasingly large and complex data sets – often called
“Big Data”. The application of data mining in the
domain of bioinformatics is explained. It also highlights
some of the current challenges and opportunities of data
mining in bioinformatics.
Keywords: Data Mining, HCI and KDD (Human (Human-
Computer Interaction, Knowledge Discovery & Data
Mining), Big Data, Interactive Knowledge discovery.
Introduction
Data mining is defined as the process of automatically
extracting meaningful patterns from usually very large
quantities of seeminglyly unrelated data. Data mining
emerges as a new discipline at the end of 1980’s. The
introduction of new technologies such as computers,
satellites, new mass storage media and many others
have leads to an exponential growth of collected data.
Traditional data
ata analysis techniques often failed to
process large amounts of often noisy data efficiently, in
an exploratory fashion. The scope of data mining is the
knowledge extraction from large data amounts with the
help of computers. It is an interdisciplinary areas
ar of
researches that has its roots in databases, machine
learning and statistics and has contribution from many
other areas such as information retrieval, pattern
recognition, visualization, parallel and distributed
computing. It is an iterative process in which preceding
process are modified to support new hypotheses
suggested by the data. The main aim of data mining is
to explore the databases through automated means and
discover meaningful, useful patterns and relationships
in data. Data mining can be defined as one particular
step of the KDD (knowledge discovery from data)
process: the identification of interesting structures in
data. It uses different algorithm for classification,
regression, clustering or association rules.
The steps for data mining follow the following pattern:
 Data Extraction
 Data Cleansing
 Data Transformation /Reduction
 Data Mining Methods
 Applying Data Mining Algorithm
 Modeling Data
 Pattern Discovery
 Data Visualization
Data Extraction
Data selection and sampling from extracted data by
data warehouses, databases data marts repositories is a
first challenging step in data mining. Data mining
requires a controlled vocabulary, usually implemented
as part of a data dictionary, so that a single word can be

@ IJTSRD | Available Online @ www.ijtsrd.com | Volume – 2 | Issue – 1 | Nov-Dec

Dec 2017 Page: 1109
 International Journal of Trend in Scientific Research and Development (IJTSRD) ISSN: 2456-6470
used to express a given concept. As millions and
thousands of records and variables are gathered in data
warehouses and data bases initial mining of meaningful
data is quite a complicated process. Typically restrict to
computationally enable sample of the holding in an
entire data warehouse. The evaluation of the
relationships that are revealed in these samples can be
used to determine which relationships in the data
should be mined further using the complete data
warehouse. With large, complex databases, even with
sampling, the computational resource requirements
associated with non-directed data mining may be
excessive. In this situation, researchers generally rely
on their knowledge of biology to identify potentially
valuable relationships and they limit sampling based on
these heuristics.
Data Cleansing
The data collected are not clean and may contain errors,
missing values, noisy or inconsistent data. So we need
to apply different techniques to get rid of such
anomalies.
Once this extracted it has to be preprocessed and
cleaned. This is done in following steps:
Data Characterization: It basically deals with
documentation of data in an appropriate and
meaningful manner, so that any person could
understand and interpret the data comfortably. This task
is basically done by programmers and other staff
involved in data mining project it involves creating a
high-level description of the nature and the content of
the data to be mined.
Consistency Analysis: It is analyzing the variability of
data independent of domain. Based on data values, it is
primarily statistical analysis of data. Outliers and values
determined to be significantly different from other data
may be automatically excluded from the knowledge-
discovery process, based on predefined statistical
constraints.
For example, data associated with a given parameter
that is more than three standard. Deviations from the
mean might be excluded from the mining operation.
Domain Analysis: It is validating the data values in a
larger context of biology. It is something which goes
beyond simply verifying that data value is a text string
or an integer, or that it's statistically consistent with
other data on the same parameter, to ensure that it
makes sense in the context of the biology. Domain
analysis requires that someone familiar with the
@ IJTSRD | Available Online @ www.ijtsrd.com | Volume – 2 | Issue – 1 | Nov-Dec 2017
biology create the heuristics that can be applied to the
data. Data enrichment: involves strengthening of data
from multiple data sources to minimize the limitations
of a single data source. It basically involves studying
various sources of data For example; two databases on
inherited diseases might each be sparsely populated in
terms of proteins that are associated with particular
diseases. This deficit could be addressed by
incorporating data from both databases, assuming only
a moderate degree of overlap in the content of the two
databases. Frequency and Distribution Analysis: It
finds the frequency of occurrence of data during the
data mining process by placing the weights on values as
a function of their frequency of Occurrence. This is
done to maximize the contribution of common findings
while minimizing the effect of rare occurrences on the
conclusions made from the data-mining output.
Data Transformation
The data even after cleaning are not ready for mining as
we need to transform them into forms appropriate for
mining. The techniques used to accomplish this are
smoothing, aggregation, normalization etc.
Normalization: It represents the data in various forms
depending on analysis and based on further processes to
be implemented. It involves transforming data values
from one representation to another, using a predefined
range of final values. Various scales are used in
normalization process like absolute scales, nominal
scales, ordinal scales, rank scales. For example,
qualitative values, such as "high" and "low," and
qualitative values from multiple sources regarding a
particular parameter might be normalized to a
numerical score from 1 to 10.
Data Mining
Now we are ready to apply data mining techniques on
the data to discover the interesting patterns. The
process of data mining is concerned with extracting
patterns from the data. Techniques like clustering and
association analysis are among the many different
techniques used for data mining.
Applying Data Mining Algorithm
This is not a single method or approach, but ii
converges various technology and techniques to
achieve proper mining of wide range of and also the
data of interest biological data. Machine learning
methods have wide applicability in data mining
algorithms. It includes statistics, biological modeling,
adaptive control theory, psychology, and artificial
Page: 1110
 International Journal of Trend in Scientific Research and Development (IJTSRD) ISSN: 2456-6470
intelligence (AI).Basically genetic algorithm and neural
networks take a major part as a technique to in
biological data. Similarly, adaptive control theory,
where parameters of System change dynamically to
meet the current conditions, and psychological theories,
especially those regarding positive and negative
reinforcement learning, heavily influence machine
learning methods. Artificial Intelligence techniques,
such as pattern matching through inductive logic
programming, are designed to derive general rules from
specific examples.
Data Modeling
Data modeling basically is a process of structuring and
organizing the data, and then these structured data are
implemented in database management system. Today’s
biological world demands for heavy exploitation of
data These data as are in various forms which has to be
capsulated in a meaning full manner .The data are in
disparate formats, remotely dispersed, and based on the
different vocabularies of Various disciplines.
Furthermore, data are often stored or distributed using
formats that leave implicit many important features
relating to the structure and semantics of the data.
Conceptual data modeling involves the development of
implementation-independent models that capture and
make explicit the principal structural properties of data.
Entities such as a biopolymer or a reaction, and their
relations, egg catalyzed can be formalized using a
conceptual data model. Conceptual models are
implementation-independent and can be transformed in
systematic ways for implementation using different
platforms, e.g. traditional database management
systems.
Pattern Discovery
Biology has been transformed from a data poor to a
data rich field, with massive accumulation of disparate
types of data, for example huge databases of sequences
(DNA, RNA, or protein). This data allows important
biological insights to be made, partly by finding
patterns and motifs that are conserved across many
individuals or species; there is now a huge biological
literature reporting on such conserved patterns and
motifs that have been found in biological datasets. In
contrast to the area of pattern matching, the patterns
and motifs are generally not known ahead of time, but
must be identified or discovered from the data; this task
is often very subtle and difficult because the patterns
and motifs may be short, may be highly degenerate
(containing wildcards and variable length elements),
may be ordered differently in different genomes, and
@ IJTSRD | Available Online @ www.ijtsrd.com | Volume – 2 | Issue – 1 | Nov-Dec 2017
are generally hidden in that they make up a small
fraction of the data. For particular biological
applications, even the definition of a relevant pattern
may be difficult to state clearly, or may be unresolved.
In bioinformatics, pattern recognition is most often
concerned with the automatic classification of character
sequences representative of the nucleotide bases or
molecular structures, and of 3D protein structures.
Data Visualization
Visualizing biological data is one of the most
challenging part of data mining process. In this modern,
digital society, how the data is visualized becomes the
prime facto, when it comes to communicating or
understanding complex concepts. Better the data
visualized, better the concepts will be clear.
Visualization technologies can provide an intuitive
representation of the relationships among large groups
of objects or data points that could otherwise be
incomprehensible, while providing context and
indications of relative importance. The "Sequence
Visualization" and "Structure Visualization is types of
data visualization techniques.
Houle et al. (2000) refer to a classification of three
successive levels for the analysis of biological data,
that is identified on the basis of the central dogma of
molecular biology:
1. Genomics is the study of an organism's genome and
deals with the systematic use of genome information to
provide new biological knowledge.
2. Gene expression analysis is the use of quantitative
mRNA-level measurements of gene expression (the
process by which a gene’s coded information is
converted into the structural and functional units of a
cell) in order to characterize biological processes and
elucidate the mechanisms of gene transcription (Houle
et al., 2000).
3. Proteomics is the large-scale study of proteins,
particularly their structures and functions. These
application domains are examined in the following
paragraphs. As many genome projects (the endeavors to
sequence and map genomes) like the Human Genome
Project have been completed, there is a paradigm shift
from static structural genomics to dynamic functional
genomics (Houle et al., 2000). The term structural
genomics refers to the DNA sequence determination
and mapping activities, while functional genomics
refers to the assignment of functional information to
known sequences. There are particular DNA sequences
Page: 1111
 International Journal of Trend in Scientific Research and Development (IJTSRD) ISSN: 2456-6470
that have a specific biological role. The identification
of such sequences is a problem that concerns
bioinformatics scientists. One such sequence is
transcription start site, which is the region of DNA
where transcription (the process of mRNA production
from DNA) starts. Another biologically meaningful
sequence is the translation initiation site, which is the
site where translation (protein production from mRNA)
initiates. Although every cell in an organism -with only
few exceptions- has the same set of chromosomes, two
cells may have very different properties and functions.
This is due to the differences in abundance of proteins.
The abundance of a protein is partly determined by the
levels of mRNA which in turn are determined by the
expression or non-expression of the corresponding
gene. A tool for analyzing gene expression is
microarray. A microarray experiment measures the
relative mRNA levels of typically thousands of genes,
providing the ability to compare the expression levels
of different biological samples. These samples may
correlate with different time points taken during a
biological process or with different tissue types such as
normal cells and cancer cells (Aas, 2001).
Serial Analysis of Gene Expression (SAGE) is a
method that allows the quantitative profiling of a large
number of transcripts (Velculescu et al., 1995). A
transcript is a sequence of mRNA produced by
transcription. However, this method is very expensive
in contrast to microarrays, thus there is a limited
amount of publicly available SAGE data. One of the
concerns of Proteomics is the prediction of protein
properties such as active sites, modification sites,
localization, stability, globularity, shape, protein
domains, secondary structure and interactions
(Whishart, 2002). Secondary structure prediction is one
of the most important problems in proteomics. The
interaction of proteins with other biomolecules is
another important issue.
Mining Biological Data
Data mining is the discovery of useful knowledge from
databases. It is the main step in the process known as
Knowledge Discovery in Databases (KDD) (Fayyad et
al., 1996), although the two terms are often used
interchangeably. Other steps of the KDD process are
the collection, selection, and transformation of the data
and the visualization and evaluation of the extracted
knowledge. Data mining employs algorithms and
techniques from statistics, machine learning, artificial
intelligence, databases and data warehousing etc. Some
of the most popular tasks are classification, clustering,
@ IJTSRD | Available Online @ www.ijtsrd.com | Volume – 2 | Issue – 1 | Nov-Dec 2017
association and sequence analysis, and regression.
Depending on the nature of the data as well as the
desired knowledge there is a large number of
algorithms for each task. All of these algorithms try to
fit a model to the data (Dunham, 2002). Such a model
can be either predictive or descriptive. A predictive
model makes a prediction about data using known
examples, while a descriptive model identifies patterns
or relationships in data. Table 3 presents the most
common data mining tasks (Dunham, 2002).
Many general data mining systems such as SAS
Enterprise Miner, SPSS, S-Plus, IBM Intelligent Miner,
Microsoft SQL Server 2000, SGI MineSet, and
InxightVizServer can be used for biological data
mining. However, some biological datamining tools
such as GeneSpring, Spot Fire, VectorNTI, COMPASS,
Statistics for Microarray Analysis, and Affymetrix Data
Mining Tool have been developed (Han, 2002). Also, a
large number of biological data mining tools is
provided by National Center for Biotechnology
Information and by European Bioinformatics Institute.
Data Mining in Genomics
Many data mining techniques have been proposed to
deal with the identification of specific DNA sequences.
The most common include neural networks, Bayesian
classifiers, decision trees, and Support Vector Machines
(SVMs) (Ma & Wang, 1999; Hirsh &Noordewier,
1994; Zien et al., 2000). Sequence recognition
algorithms exhibit performance tradeoffs between
increasing sensitivity (ability to detect true positives)
and decreasing selectivity (ability to exclude false
positives) (Houle et al., 2000). However, as Li et al.
(2003) state, traditional data mining techniques cannot
be directly applied to this type of recognition problems.
Thus, there is the need to adapt the existing techniques
to this kind of problems. Attempts to overcome this
problem have been made using feature generation and
feature selection (Zeng& Yap, 2002; Li et al., 2003).
Another data mining application in genomic level is the
use of clustering algorithms to group structurally
related DNA sequences.
Gene Expression Data Mining
The main types of microarray data analysis include
(Piatetsky-Shapiro & Tamayo,2003): gene selection,
clustering, and classification. Piatetsky-Shapiro and
Tamayo (2003) present one great challenge that data
mining practitioners have to deal with. Microarray
datasets -in contrast with other application domains-
contain a small number of records (less than a
Page: 1112
 International Journal of Trend in Scientific Research and Development (IJTSRD) ISSN: 2456-6470
hundred), while the number of fields (genes), is
typically in thousands. The same case is in SAGE data.
This increases the likelihood of finding “false
positives”. An important issue in data analysis is
feature selection. In gene expression analysis the
features are the genes. Gene selection is a process of
finding the genes most strongly related to a particular
class. One benefit provided by this process is the
reduction of the foresaid dimensionality of dataset.
Moreover, a large number of genes are irrelevant when
classification is applied. The danger of overshadowing
the contribution of relevant genes is reduced when gene
selection is applied. Clustering is the far most used
method in gene expression analysis. Tibshirani et al.
(1999) and Aas (2001) provide a classification of
clustering methods in two categories: one-way
clustering and two-way clustering. Methods of the first
category are used to group either genes with similar
behavior or samples with similar gene expressions.
Two-way clustering methods are used to
simultaneously cluster genes and samples. Hierarchical
clustering is currently the most frequently applied
method in gene expression analysis. An important issue
concerning the application of clustering methods in
microarray data is the assessment of cluster quality.
Many techniques such as bootstrap, repeated
measurements, mixture model-based approaches, sub-
sampling and others have been proposed to deal with
the cluster reliability assessment (Kerr &Churchill,
2001; Yeung et al., 2003; Ghosh&Chinnaiyan, 2002;
Smolkin&Ghosh, 2003). In microarray analysis
classification is applied to discriminate diseases or to
predict outcomes based on gene expression patterns and
perhaps even identify the best treatment for given
genetic signature (Piatetsky-Shapiro & Tamayo, 2003).
Table 4 lists the most commonly used methods in
microarray data analysis. Detailed descriptions of these
methods can be found in literature (Aas, 2001;
Tibshirani et al., 1999; Hastie et al., 2000; Lazzeroni&
Owen, 2002; Dudoit et al., 2002; Golub et al., 1999).
Most of the methods used to deal with microarray data
analysis can be used for SAGE data analysis. Finally,
machine learning and data mining can be applied in
order to design microarray experiments except to
analyze them (Molla et al., 2004).
Data Mining in Proteomics
Many modification sites can be detected by simply
scanning a database that contains known modification
sites. However, in some cases, a simple database scan is
not effective. The use of neural networks provides
@ IJTSRD | Available Online @ www.ijtsrd.com | Volume – 2 | Issue – 1 | Nov-Dec 2017
better results in these cases. Similar approaches are
used for the prediction of active sites. Neural network
approaches and nearest neighbor classifiers have been
used to deal with protein localization prediction
(Whishart, 2002). Neural networks have also been used
to predict protein properties such as stability,
globularity and shape. Whishart refers to the use of
hierarchical clustering algorithms for predicting protein
domains. Data mining has been applied for the protein
secondary structure prediction. This problem has been
studied for over than 30 years and many techniques
have been developed (Whishart, 2002). Initially,
statistical approaches were adopted to deal with this
problem. Later, more accurate techniques based on
information theory, Bayes theory, nearest neighbors,
and neural networks were developed. Combined
methods such as integrated multiple sequence
alignments with neural network or nearest neighbor
approaches improve prediction accuracy. A density
based clustering algorithm (GDBSCAN) is presented
by Sander et al. (1998), that can be used to deal with
protein interactions. This algorithm is able to cluster
point and spatial objects according to both, their spatial
and non-spatial attributes.
Databases of Bioinformatics
There are many rapidly growing databases in the field
of Bio informatics.
Protein Data Bank
The PDB archive is a repository of atomic coordinates
and other information describing proteins and other
important biological macromolecules. Structural
biologists use methods such as X-ray crystallography,
NMR spectroscopy, and cryo-electron microscopy to
determine the location of each atom relative to each
other in the molecule. They then deposit this
information, which is then annotated and publicly
released into the archive by the PDB.
SWISS-PROT:
SWISS-PROT is an annotated protein sequence
database, which was created at the Department of
Medical Biochemistry of the University of Geneva and
has been a collaborative effort of the Department and
the European Molecular Biology Laboratory (EMBL),
since 1987. The SWISS-PROT protein sequence
database consists of sequence entries. Sequence entries
are composed of different line types, each with their
own format. For standardization purposes the format of
SWISS-PROT follows as closely as possible that of the
Page: 1113
 International Journal of Trend in Scientific Research and Development (IJTSRD) ISSN: 2456-6470
EMBL Nucleotide Sequence Database. The SWISS-
PROT database distinguishes itself from other protein
sequence databases by three distinct criteria:
(i) Annotations,
(ii) Minimal redundancy and
(iii) Integration with other databases.
Medline
Medical Literature Analysis and Retrieval System
Online, or MEDLARS Online is a bibliographic
database of life sciences and biomedical information. It
includes bibliographic information for articles from
academic Journals covering medicine, nursing,
pharmacy, dentistry, veterinary medicine, and health
care. MEDLINE also covers much of the literature in
biology and biochemistry, as well as fields such as
molecular evolution.
The EMBL Nucleotide Sequence Database
The EMBL Database collects, organizes and distributes
a database of nucleotide sequence data and related
biological information. Since 1982 this work has been
done in collaboration with GenBank (NCBI, Bethesda,
USA) and the DNA Database of Japan (Mishima). Each
of the three international collaborating databases
DDBJ/EMBL/GenBank, collect a portion of the total
sequence data reported world-wide. All new and
updated database entries are exchanged between the
International Nucleotide Sequence Collaboration on a
daily basis. EMBL Database releases are produced
quarterly and are distributed on CD-ROM. The most
up-to-date data collection is available via Internet and
World Wide Web interface.
Biological Data Analysis
Biological data mining is a very important part of
Bioinformatics. Following are the aspects in which data
mining contributes for biological data analysis:
 Semantic integration of heterogeneous, distributed
genomic and proteomic databases.
 Alignment, indexing, similarity search and
comparative analysis multiple nucleotide
sequences.
@ IJTSRD | Available Online @ www.ijtsrd.com | Volume – 2 | Issue – 1 | Nov-Dec 2017
 Discovery of structural patterns and analysis of
genetic networks and protein pathways.
 Association and path analysis.
 Visualization tools in genetic data analysis.
Applications of data mining in bioinformatics
 Gene finding,
 Protein function domain detection,
 Function motif detection,
 Protein function inference,
 Disease diagnosis,
 Disease prognosis,
 Disease treatment optimization,
 Protein and gene interaction network,
 Reconstruction,
 Data cleansing,
 Protein sub-cellular location prediction,
 Analysis of protein and nucleotides sequences.
REFERENCES
1. M. Andrade and P. Bork. Automated extraction of
Information in molecular biology. FEBS
Letters,476:12–17, 2000.
2. T. Attwood and D. Parry-Smith. Introduction to
Bioinformatics. Longman Higher Education, 1999.
3. A. Bairoch and R. Apweiler. The SWISS-PROT
Protein sequence database and its supplement
TrEMBL in 2000. NucleicAcids Res., 28:45–48,
2000.
4. P. Baldi and S. Brunak. Bioinformatics: The
MachineLearning Approach, Second Edition. MIT
Press, 2001.
5. Bioinformatics Computing By Bryan Bergeron.
Page: 1114